We Graded Our Own Ownership Projections — MLB Main Slate, 8/11

Twelve games, and a correction we owe you from yesterday that changes how every number below should be read.

Yesterday we published a playbook whose entire opening argument was that Baltimore was the most overpriced stack on the board at 57.6% projected ownership. Then the contest settled. Baltimore's most-owned bat finished at 15.9%. Gunnar Henderson, projected 16.8%, came in at 11.5%. Jackson Holliday, projected 16.5%, came in at 11.9%.

The read was directionally fine — Baltimore was overvalued and they scored one run. But the number the argument rested on was wrong by a factor of three, and we're not going to pretend otherwise.

So we pulled the settled standings from a 4,004-entry contest and graded every ownership projection on the slate against what the field actually did. Then we did the same on a 1,176-entry WNBA contest. Two slates, 5,180 real lineups. Here is what came back.

Ownership projections are wrong in a specific, repeatable direction

The pattern is consistent and it runs both ways. Mid-priced value bats get more owned than projected — often twice as much. The field finds them; they are not the secret the projection implies. And concentrated team chalk gets less owned than projected, because ownership spreads across a roster rather than piling into the four names a model flags.

The WNBA slate showed the same thing in a different shape. Angel Reese projected 34.9% and finished at 59.1% — the field concentrated into the single obvious stud far harder than the model expected. Brionna Jones projected 43.8% and finished at 24.3%. Mean absolute error across that slate: 6.1 percentage points.

A leverage play built on a 4% projection that is really 12% is not a leverage play. It is a normal play you paid a story for.

What we changed

Three things, starting with tonight's board:

1. We no longer quote a single ownership number as fact. Anything under roughly 8% projected should be read as a range, and the top of that range is often double the printed figure.
2. Team-level ownership sums are treated as soft. A "67% owned stack" almost never means 67% of lineups have that stack. It means the five bats sum to 67 across a field that is spreading its exposure.
3. We rank stacks on ceiling first and ownership second, rather than leading with the ownership-per-ceiling ratio, because the denominator is the least reliable number in the build.

Tonight: twelve games

The board is wider than yesterday's — twelve games, twenty-four teams — and the total range runs from 7.0 up to the 10.5 in Sacramento. Here is the full stack board, sorted by the summed ceiling of each team's top five batting-order slots.

Colorado tops the ceiling board for the second night running — 87.0 against Mitch Bratt in Chase Field, at 16.7% projected. Read that 16.7% as our new rules demand: soft, probably higher, still not chalk.

Below the fold: the chalk autopsy on Tampa, Baltimore and Arizona, all sixteen projected starters graded, the arm at 2.0% ownership with the highest leverage score on the board, the one-off bats, a game-by-game read, and six cap-validated DK builds.